Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
THE KEYNESIAN THEORY AND THE GEOGRAPHIC CONCENTRATION IN THE PORTUGUESE MANUFACTURED INDUSTRY
1. THE KEYNESIAN THEORY AND THE GEOGRAPHIC
CONCENTRATION IN THE PORTUGUESE MANUFACTURED
INDUSTRY
Vitor João Pereira Domingues Martinho
Unidade de I&D do Instituto Politécnico de Viseu
Av. Cor. José Maria Vale de Andrade
Campus Politécnico
3504 - 510 Viseu
(PORTUGAL)
e-mail: vdmartinho@esav.ipv.pt
ABSTRACT
This work aims to test the Verdoorn Law, with the alternative specifications of (1)Kaldor (1966),
for the five Portuguese regions (NUTS II), from 1986 to 1994. It is intended to test, yet in this work, the
alternative interpretation of (2)Rowthorn (1975) about the Verdoorn's Law for the same regions and period.
The results of this study are about each one of the manufactured industries operating in the Portuguese
regions. This paper pretends, also, to analyze the importance which the natural advantages and local
resources are in the manufacturing industry location, in relation with the "spillovers" effects and industrial
policies. To this, we estimate the Rybczynski equation matrix for the various manufacturing industries in
Portugal, at regional level (NUTS II) and for the period 1986 to 1994.
Keywords: Verdoorn law; geographic concentration; panel data; manufactured industries;
Portuguese regions.
1. INTRODUCTION
Kaldor rediscovered the Verdoorn law in 1966 and since then this law has been tested in several
ways, using specifications, samples and different periods (3)(Martinho, 2011). However, the conclusions
drawn differ, some of them rejecting the Law of Verdoorn and other supporting its validity. (4)Kaldor (1966,
1967) in his attempt to explain the causes of the low rate of growth in the UK, reconsidering and
empirically investigating Verdoorn's Law, found that there is a strong positive relationship between the
growth of labor productivity (p) and output (q), i.e. p = f (q). Or alternatively between employment growth
(e) and the growth of output, ie, e = f (q).
Another interpretation of Verdoorn's Law, as an alternative to the Kaldor, is presented by
(5)Rowthorn (1975, 1979). Rowthorn argues that the most appropriate specification of Verdoorn's Law is
the ratio of growth of output (q) and the growth of labor productivity (p) with employment growth (e), i.e., q
= f (e) and p = f (e), respectively (as noted above, the exogenous variable in this case is employment). On
the other hand, Rowthorn believes that the empirical work of Kaldor (1966) for the period 1953-54 to 1963-
64 and the (6)Cripps and Tarling (1973) for the period 1951 to 1965 that confirm Kaldor's Law, not can be
accepted since they are based on small samples of countries, where extreme cases end up like Japan
have great influence on overall results.
Taking into account the work of (7)Kim (1999), we seek, aldo, to analyze the importance of the
natural advantages and local resources (specific factors of locations) have in explaining the geographic
concentration over time in the Portuguese regions, relatively effects "spillovers" and industrial policies (in
particular, the modernization and innovation that have allowed manufacturing in other countries take better
advantage of positive externalities). For this, we estimated the Rybczynski equation matrix for the different
manufacturing industries in the regions of Portugal, for the period 1986 to 1994. It should be noted that
while the model of inter-regional trade, the Heckscher-Ohlin-Vanek, presents a linear relationship between
net exports and inter-regional specific factors of locations, the Rybczynski theorem provides a linear
relationship between regional production and specific factors of locations. In principle, the residual part of
the estimation of Rybczynski, measured by the difference between the adjusted degree of explanation
(R2) and the unit presents a approximated estimate of the importance not only of the "spillovers” effects,
as considered by Kim (1999), but also of the industrial policies, because, industrial policies of
modernization and innovation are interconnected with the "spillover" effects. However, it must be some
caution with this interpretation, because, for example, although the growth of unexplained variation can be
attributed to the growing importance of externalities "Marshallians" or "spillovers" effects and industrial
policies, this conclusion may not be correct. Since the "spillovers" effects and industrial policies are
measured as a residual part, the growth in the residual can be caused, also, for example, by growth in the
randomness of the location of the products manufactured and the growing importance of external trade in
goods and factors.
1
2. 2. ALTERNATIVE SPECIFICATIONS OF VERDOORN'S LAW
The hypothesis of increasing returns to scale in industry was initially tested by Kaldor (1966)
using the following relations:
pi a bqi , Verdoorn law (1)
ei c dqi , Kaldor law (2)
where pi, qi and ei are the growth rates of labor productivity, output and employment in the industrial
sector in the economy i.
On the other hand, the mathematical form of Rowthorn specification is as follows:
pi 1 1ei , firts equation of Rowthorn (3)
qi 2 2 ei , second equation of Rowthorn (4)
where 1 2 e 2 (1 1 ) , because pi=qi-ei. In other words, qi ei 1 1ei ,
qi 1 ei 1ei , so, qi 1 (1 1 )ei .
Rowthorn estimated these equations for the same OECD countries considered by Kaldor (1966),
with the exception of Japan, and for the same period and found that 2 was not statistically different from
unity and therefore 1 was not statistically different from zero. This author thus confirmed the hypothesis
of constant returns to scale in manufacturing in the developed countries of the OECD. (8)Thirlwall (1980)
criticized these results, considering that the Rowthorn interpretation of Verdoorn's Law is static, since it
assumes that the Verdoorn coefficient depends solely on the partial elasticity of output with respect to
employment.
3. THE MODEL THAT ANALYZES THE IMPORTANCE OF NATURAL ADVANTAGES AND
LOCAL RESOURCES IN AGGLOMERATION
According to Kim (1999), the Rybczynski theorem states that an increase in the supply of one
factor leads to an increased production of the good that uses this factor intensively and a reduction in the
production of other goods.
Given these assumptions, the linear relationship between regional output and offers of regional
factors, may be the following:
Y A1V ,
where Y (nx1) is a vector of output, A (nxm) is a matrix of factor intensities or matrix input Rybczynski and
V (mx1) is a vector of specific factors to locations.
For the output we used the gross value added of different manufacturing industries, to the specific
factors of the locations used the labor, land and capital. For the labor we used the employees in
manufacturing industries considered (symbolized in the following equation by "Labor") and the capital,
because the lack of statistical data, it was considered, as a "proxy", the production in construction and
public works (the choice of this variable is related to several reasons including the fact that it represents a
part of the investment made during this period and symbolize the part of existing local resources,
particularly in terms of infrastructure). With regard to land, although this factor is often used as specific of
the locations, the amount of land is unlikely to serve as a significant specific factor of the locations.
Alternatively, in this work is used the production of various extractive sectors, such as a "proxy" for the
land. These sectors, include agriculture, forestry and fisheries (represented by "Agriculture") and
production of natural resources and energy (symbolized by "Energy"). The overall regression is then used
as follows:
ln Yit 1 ln Laborit 2 ln Agricultur eit 3 ln Energy it 4 ln Constructi onit
In this context, it is expected that there is, above all, a positive relationship between the
production of each of the manufacturing industry located in a region and that region-specific factors
required for this industry, in particular, to emphasize the more noticeable cases, between food industry and
agriculture, among the textile industry and labor (given the characteristics of this industry), among the
industry of metal products and metal and mineral extraction and from the paper industry and forest.
2
3. 4. DATA ANALYSIS
Considering the variables on the models presented previously and the availability of statistical
information, we used the following data disaggregated at regional level. Annual data for the period 1986 to
1994, corresponding to the five regions of mainland Portugal (NUTS II), and for the several manufactured
industries in those regions. The data are relative, also, to regional gross value added of agriculture,
fisheries and forestry, natural resources and energy and construction and public works. These data were
obtained from Eurostat (Eurostat Regio of Statistics 2000).
5. EMPIRICAL EVIDENCE OF THE VERDOORN'S LAW
The results in Table 1, obtained in the estimations carried out with the equations of Verdoorn,
Kaldor and Rowthorn for each of the manufacturing industries, enable us to present the conclusions
referred following.
Manufacturing industries that have, respectively, higher increasing returns to scale are the
industry of transport equipment (5.525), the food industry (4.274), industrial minerals (3.906), the metal
industry (3.257), the several industry (2.222), the textile industry (1.770), the chemical industry (1.718) and
industry equipment and electrical goods (presents unacceptable values). The paper industry has
excessively high values. Note that, as expected, the transportation equipment industry and the food
industry have the best economies of scale (they are modernized industries) and the textile industry has the
lowest economies of scale (industry still very traditional, labor intensive, and in small units).
Also in Table 1 presents the results of an estimation carried out with 9 manufacturing industries
disaggregated and together (with 405 observations). By analyzing these data it appears that were obtained
respectively for the coefficients of the four equations, the following elasticities: 0.608, 0.392, -0.275 and
0.725. Therefore, values that do not indicate very strong increasing returns to scale, as in previous
estimates, but are close to those obtained by Verdoorn and Kaldor.
Table 1: Analysis of economies of scale through the equation Verdoorn, Kaldor and Rowthorn, for each of
the manufacturing industries and in the five NUTS II of Portugal, for the period 1986 to 1994
Metal Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
Verdoorn -4.019* 0.693*
pi a bqi
1.955 0.898 29
(-2.502) (9.915)
Kaldor 4.019* 0.307*
ei c dqi 1.955 0.788 29
(2.502) (4.385)
Rowthorn1 3.257
-12.019 0.357
pi 1 1ei 1.798 0.730 29
(-0.549) (1.284)
Rowthorn2 -12.019 1.357*
qi 2 2 ei 1.798 0.751 29
(-0.549) (4.879)
Mineral Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.056* 0.744*
Verdoorn 1.978 0.352 38
(-4.296) (4.545)
0.056* 0.256
Kaldor 1.978 0.061 38
(4.296) (1.566)
3.906
-0.023 -0.898*
Rowthorn1 2.352 0.704 38
(-0.685) (-9.503)
-0.023 0.102
Rowthorn2 2.352 0.030 38
(-0.685) (1.075)
Chemical Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
0.002 0.418*
Verdoorn 1.825 0.554 34
(0.127) (6.502)
-0.002 0.582*
Kaldor 1.825 0.707 34
(-0.127) (9.052)
1.718
9.413* 0.109
Rowthorn1 1.857 0.235 33
(9.884) (0.999)
9.413* 1.109*
Rowthorn2 1.857 0.868 33
(9.884) (10.182)
Electrical Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
0.004 -0.126
Verdoorn 1.762 0.128 32
(0.208) (-1.274)
-0.004 1.126*
Kaldor 1.762 0.796 32
(-0.208) (11.418)
---
0.019 -0.287*
Rowthorn1 1.659 0.452 32
(1.379) (-4.593)
0.019 0.713*
Rowthorn2 1.659 0.795 32
(1.379) (11.404)
Transport Industry
3
4. Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.055* 0.819*
Verdoorn 2.006 0.456 38
(-2.595) (5.644)
0.055* 0.181
Kaldor 2.006 0.040 38
(2.595) (1.251)
5.525
-0.001 -0.628*
Rowthorn1 2.120 0.436 32
(-0.029) (-3.938)
-0.001 0.372*
Rowthorn2 2.120 0.156 32
(-0.029) (2.336)
Food Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
0.006 0.766*
Verdoorn 2.191 0.526 38
(0.692) (6.497)
-0.006 0.234**
Kaldor 2.191 0.094 38
(-0.692) (1.984)
4.274
0.048* -0.679*
Rowthorn1 1.704 0.324 38
(2.591) (-4.266)
0.048* 0.321*
Rowthorn2 1.704 0.097 38
(2.591) (2.018)
Textile Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.008 0.435*
Verdoorn 2.117 0.271 34
(-0.466) (3.557)
0.008 0.565*
Kaldor 2.117 0.386 34
(0.466) (4.626)
1.770
0.002 -0.303*
Rowthorn1 1.937 0.136 34
(0.064) (-2.311)
0.002 0.697*
Rowthorn2 1.937 0.454 34
(0.064) (5.318)
Paper Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.062* 1.114*
Verdoorn 1.837 0.796 38
(-3.981) (12.172)
0.062* -0.114
Kaldor 1.837 0.039 38
(3.981) (-1.249)
0.028 -1.053*
Rowthorn1 1.637 0.310 38
(1.377) (-4.134)
0.028 -0.053
Rowthorn2 1.637 0.001 38
(1.377) (-0.208)
Several Industry
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-1.212 0.550*
Verdoorn 2.185 0.529 37
(-0.756) (8.168)
1.212 0.450*
Kaldor 2.185 0.983 37
(0.756) (6.693)
2.222
8.483* 0.069
Rowthorn1 2.034 0.175 37
(24.757) (1.878)
8.483* 1.069*
Rowthorn2 2.034 0.975 37
(24.757) (29.070)
9 Manufactured Industry Together
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.030* 0.608*
Verdoorn 1.831 0.516 342
(-6.413) (19.101)
0.030* 0.392*
Kaldor 1.831 0.308 342
(6.413) (12.335)
2.551
-0.003 -0.275*
Rowthorn1 1.968 0.053 342
(-0.257) (-4.377)
-0.003 0.725*
Rowthorn2 1.968 0.280 342
(-0.257) (11.526)
Note: * Coefficient statistically significant at 5%, ** Coefficient statistically significant at 10%, GL,
Degrees of freedom; EE, Economies of scale.
6. EMPIRICAL EVIDENCE OF GEOGRAPHIC CONCENTRATION
In the results presented in the following table, there is a strong positive relationship between
gross value added and labor in particular in the industries of metals, chemicals, equipment and electrical
goods, textile and several products. On the other hand, there is an increased dependence on natural and
local resources in industries as the mineral products, equipment and electric goods, textile and several
products. We found that the location of manufacturing industry is yet mostly explained by specific factors of
locations and poorly explained by "spillovers" effects and industrial policies.
4
5. Table 2: Results of estimations for the years 1986-1994
ln Yit 1 ln Laborit 2 ln Agricultur eit 3 ln Energy it 4 ln Constructi onit
IMT IMI IPQ IEE IET IAL ITE IPA IPD
(2) (1) (1) (1) (1) (2) (1) (1) (2)
10.010 34.31(*) 83.250(*)
(0.810) (3.356) (5.412)
Dummy1 18.753(*) -13.467(*) 14.333(*) 9.183 15.175(*) 17.850(*)
(5.442) (-3.134) (2.811) (1.603) (3.652) (3.162)
Dummy2 19.334(*) -12.679(*) 13.993(*) 10.084(**) 14.904(*) 17.532(*)
(5.733) (-2.930) (2.802) (1.766) (3.597) (3.100)
Dummy3 19.324(*) -13.134(*) 14.314(*) 10.155(**) 14.640(*) 18.586(*)
(5.634) (-3.108) (2.804) (1.797) (3.534) (3.313)
Dummy4 18.619(*) -11.256(*) 14.022(*) 9.384 15.067(*) 15.001(*)
(5.655) (-2.599) (2.857) (1.627) (3.647) (2.654)
Dummy5 17.860(*) -11.060(*) 12.629(*) 7.604 13.206(*) 13.696(*)
(5.629) (-2.682) (2.653) (1.377) (3.344) (2.574)
1 1.420(*)
(4.965)
0.517(*)
(4.651)
1.098(*)
(8.056)
0.817(*)
(7.695)
0.397(*)
(2.455)
0.378(*)
(2.000)
0.809(*)
(5.962)
-0.071
(-0.230)
0.862(*)
(10.995)
2 0.844
(1.353)
-0.358(*)
(-2.420)
0.709(*)
(2.628)
-0.085
(-0.480)
-0.314
(-0.955)
-0.026
(-
-0.484(**)
(-1.952)
-0.171
(-0.505)
-0.148
(-0.780)
0.130)
3 0.431 -0.242(*)
(1.468) (-3.422)
0.120
(0.721)
-0.084
(-0.876)
0.147
(0.844)
-0.067
(-0.706)
-0.229(**)
(-1.738)
-0.165
(-0.904)
-0.524(*)
(-5.289)
4 -1.459(*) 0.359(*)
(-4.033) (2.629)
0.260
(1.185)
0.061
(0.318)
0.433(*)
(2.066)
0.166
(0.853)
0.529(*)
(2.702)
0.427
(1.596)
-0.085
(-0.461)
Sum of the 1.236 0.276 2.187 0.709 0.663 0.451 0.625 0.020 0.105
elasticities
2
R adjusted 0.822 0.993 0.987 0.996 0.986 0.968 0.997 0.983 0.999
Residual 0.178 0.007 0.013 0.004 0.014 0.032 0.003 0.017 0.001
part
Durbin- 1.901 2.246 1.624 1.538 2.137 1.513 2.318 1.956 2.227
Watson
Hausman (c) 115.873(b)(*) 26.702(b)(*) 34.002(b)(*) 9.710(b)(*) (c) 34.595(b)(*) 26.591(b)(*) 1.083(a)
test
For each of the industries, the first values correspond to the coefficients of each of the variables and values in
brackets represent t-statistic of each; (1) Estimation with variables "dummies"; (2) Estimation with random effects; (*)
coefficient statistically significant at 5% (**) Coefficient statistically significant at 10%; IMT, metals industries; IMI,
industrial mineral;, IPQ, the chemicals industries; IEE, equipment and electrical goods industries; EIT, transport
equipment industry; ITB, food industry; ITE, textiles industries; IPA, paper industry; IPD, manufacturing of various
products; (a) accepted the hypothesis of random effects; (b) reject the hypothesis of random effects; (c) Amount not
statistically acceptable.
7. CONCLUSIONS
At the level of estimates made for manufactured industries, it appears that those with,
respectively, higher dynamics are the transport equipment industry, food industry, minerals industrial,
metals industry, the several industries, the textile industry, chemical industry and equipment and electrical
goods industry. The paper industry has excessively high values.
Of referring that the location of the Portuguese manufacturing industry is still mostly explained by
specific factors of locations and the industrial policies of modernization and innovation are not relevant,
especially those that have come from the European Union, what is more worrying.
So, we can say that, although, the strong increasing returns to scale in the same industries, the
location of the manufactured industries in Portugal is mostly explained by the specific factors of the
locations.
8. REFERENCES
1. N. Kaldor. Causes of the Slow Rate of Economics of the UK. An Inaugural Lecture. Cambridge:
Cambridge University Press, 1966.
2. R.E. Rowthorn. What Remains of Kaldor Laws? Economic Journal, 85, 10-19 (1975).
3. V.J.P.D. Martinho. The Verdoorn law in the Portuguese regions: a panel data analysis. MPRA Paper
32186, University Library of Munich, Germany (2011).
4. N. Kaldor. Strategic factors in economic development. Cornell University, Itaca, 1967.
5. R.E. Rowthorn. A note on Verdoorn´s Law. Economic Journal, Vol. 89, pp: 131-133 (1979).
6. T.F. Cripps and R.J. Tarling. Growth in advanced capitalist economies: 1950-1970. University of
Cambridge, Department of Applied Economics, Occasional Paper 40, 1973.
7. S. Kim. Regions, resources, and economic geography: Sources of U.S. regional comparative
advantage, 1880-1987. Regional Science and Urban Economics (29), 1-32 (1999).
8. A.P. Thirlwall. Regional Problems are “Balance-of-Payments” Problems. Regional Studies, Vol. 14, 419-
425 (1980).
5